End to end response time too much pessimistic

[M. Zanconi]

The use of 'end to end response time' option is too much pessimistic:

For instance,

T1(2,10) --> T2(2,10) -->T3(2,10), 2 is the execution time and 10 is the period. Priorities are 1, 5 and 10 respectively. The result is :

• Compute all steps (see [2], page 3, equation 4).

Task response time

t1 => 10, Jitter= 0
t2 => 16, Jitter= 10 , missed its deadline (deadline = 10)
t3 => 12, Jitter= 10 , missed its deadline (deadline = 10)

But if I try to do this 'by hand', I see that T1 is the only one in the READY queue, so, its response time should be 2. Equation 4 gives B1=0 (since there are no tasks of lower priority than T1's) and the sum should be also 0 since there are no higher priority tasks ready to execute [T2 and T3 are not runnable]. Next T2 is the only one ready to execute, so its response time is 4 with Jitter=2 (analitically speaking B2=2 and the sum is 0). Finally T3 is read at time 4 (its jitter), finishing at 6. All tasks meet their deadline.